An Example to demonstrate the program

All files are in ASCII-text format; you can view them with
any editor. In DOS the command   more < filename is useful!

README		this file

The following files are in the used format:

train.dat	the file with the training data: 8 input varibales, one
		output variable, two classes.
		The file has 6 instances.

test.dat	the file for testing the performance of the network. The
		inputs together with the desired outputs are given.
		The file contains the 6 training cases as well as 6
		'similar' vectors.

calc.dat	the file to calculate the output of the network
		for given inputs. Only the 8 input variables are given.
		The file contains the 6 training cases as well as 6
		'similar' vectors.
		
backpro.net	this is the network description file for using the program
		bpnet.

logic.net	this is the network description file for using the program
		logicnet.


To run it:
* start the program bpnet
* select <L>oad and load the file backpro.net or <N>ew to create a network
  (for new see below); give any name for the logfile.
* select <R>eset to reset the weights (use standart parameters) 
* select <T>rain to train the network (use train.dat as data file)
  and standart parameter
* select <P>performance to evaluate the performance on the sets
  (use either train.dat or test.dat; performance on the training set
   or on the test set)
* select <C>alculate to calculate the response of the network to the
  given inputs; use the file calc.dat as input file, and calc.out 
  as output file
* quit the program

you can now see the results in the logfile. The calculated response 
is stored in calc.out


if you are using <N>ew you have to give the following answers to the program:

Setup a new Neural Network:
--- Network Parameter: --- 
Number of inputs? 8

Number of layers (2..5) [3]  : 3

Number of Neuerons in the 1. Layer? [4]2

Number of Neuerons in the 2. Layer? [4]3

Number of Neuerons in the 3. Layer = Number of outputs ? 1

Lambda value (0.1 .. 3)     [1]  : 1

Activation function (b=bi-polar, u=uni-polar)? u

Logfile for this session: logfile